In Statistics, what does p < .10 mean?

Question

This is research statistics and I'm looking for the meaning behind the symbol p < .10

Thanks

Answer 1

In statistics they use the 'rare event model' to help answer a question.

The p value is the probability that an extreme outcome of an experiment would occur if the null hypothesis was true.

If you have a very very low p value, then either what your seeing is a rare event or your null hypothesis is not correct.

For a p <0.10 is not very impressive as a rare event, ideally you would want something like p<0.05 or less.

For example, suppose some country had a space program and for $1000 you could get a ride to mars and back. You are worried that for that low a price is it safe to travel. Your null hypothesis is that it is not safe, the alternative hypothesis (if p is low ) is that it is safe to travel.

You look up in the travel journals and find that out of the 100 flights that the company has run 8 of them blew up on the launching pad, but the other flights were ok.

As a potential customer could you reject the null hypothesis that it was not safe to travel to mars with a p<0.10 ?

Answer 2

There is less than a 1 in 10 chance your event of interest occured by chance alone. Or for every 10 tests you run you will say Ah Ha when you shouldn't have at least once.

Answer 3

It means the probability of any randomly generated number falling within the range of data values specified for p is less than 1 in 10.

For example, if we have a Normal (Bell) probability density distribution for a set of data, the range of data where p < .10 resides is in the two tails of the Normal curve. The two tails are indicated by the outside range of data value beyond two standard deviations (SD) on either side of the mean average of that data. The probability of randomly generated data falling within the middle inner range of values from minus two std. dev. through plus two std. dev. would be 1 - p = .90, which is called the level of confidence.

In hypothesis testing, one of the steps before we test a hypothesis is to specify the level of confidence for the test. In your example, we can say that any measured value you get that falls outside the range indicated by the level of confidence probably did not occur randomly. That is, in all likelihood, something caused that measured value to fall outside that .90 level of confidence range.

I can't draw a Normal curve, but imagine one sitting on top of the x axis of data values shown below.

= 3SD === 2SD === 1SD ===AVE===1SD === 2SD === 3SD =.

The range of data values (x) for p < .10 comes where x >= 2SD on BOTH sides of the AVE (average) value. So if you measure something and find x = 2.1SD, for example, there is only p < .10 small chance the measured value was generated randomly. In all likelihood, something not random caused that value.

For example, suppose you measured the tolerance of a bearing inside an axle hub and found it to be x = 2.1SD. At a .90 level of confidence, that might be grounds for checking your bearing forming equipment to see if something is causing the bearing to be out of tolerance limits.

On the other hand, if x < 2SD, we might conclude the bearing tolerance was within limits and warrants no special investigation into what caused the deviation from the nominal standard specified for that bearing. This follows because that particular bearing fell withing the .90 level of confidence range of x values.

Answer 4

Probability less that 10%?

Answer 5

the probability is less than 10%

Answer 6

it means p is less than 10%

Answer 7

You "normalise" the given distribution, through calculating a corresponding z-cost. z = (X - µ) / ?, which, as you assert = -a million.0. then you definitely get a table of parts of the conventional Distribution. You seem up the tabulated cost for z = -a million.0